count-ception_mbm
count-ception_mbm copied to clipboard
In MBM class the transform parameter is not used within the `__getitem__` method so it is never applied: https://github.com/roggirg/count-ception_mbm/blob/ac0724f982a9ab0abb54e55fa8b0c171e7f3fea4/utils/dataset.py#L11.
i just cant understand last part of the code, how can i pull predictions and number of cells, and how can i test for another image after getting weights? thx
I am also curios about the estimated time to train the model with the default pickle file. I have a GTX 1080 and every Epoch is taking about 10 seconds....
I am new to deep learning and Pytorch would it be possible to use the trained model and the weights in a tensorflow model.